Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Bootstrapping inference of average treatment effect in completely randomized experiments with high-dimensional covariates

Bootstrapping inference of average treatment effect in completely randomized experiments with... Investigators often use regression adjustment methods to analyze the results of randomized experiments when baseline covariates are available. Their aim is to improve the estimation efficiency of treatment effects by adjusting for imbalance of covariates. Under mild conditions, the regression-adjusted average treatment effect estimator is asymptotically normal with asymptotic variance no greater than that of the unadjusted estimator. The asymptotic variance can be estimated conservatively based on residual sum of squares. This article studies alternative inference methods based on the bootstrap and investigates their asymptotic properties under the Neyman–Rubin causal model and randomization-based inference framework. We show that the weighted, residual and paired bootstrap methods provide asymptotically conservative variance estimators that perform at least as good as the estimator based on residual sum of squares. We further provide counterexamples, where the original estimator is asymptotically normal, but the bootstrap counterpart is inconsistent for estimating its limiting distribution. Simulation studies indicate that the paired bootstrap method is preferable, in terms of preserving type I errors, for a small sample size. Finally, our methods analyze HER2+ breast cancer data from the NeOAdjuvant Herceptin trial to examine the effectiveness of trastuzumab in combination with neoadjuvant chemotherapy. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Biostatistics & Epidemiology Taylor & Francis

Bootstrapping inference of average treatment effect in completely randomized experiments with high-dimensional covariates

Liu, Hanzhong
Biostatistics & Epidemiology , Volume OnlineFirst: 18 – Mar 17, 2021
Download PDF

Bootstrapping inference of average treatment effect in completely randomized experiments with high-dimensional covariates

Abstract

Investigators often use regression adjustment methods to analyze the results of randomized experiments when baseline covariates are available. Their aim is to improve the estimation efficiency of treatment effects by adjusting for imbalance of covariates. Under mild conditions, the regression-adjusted average treatment effect estimator is asymptotically normal with asymptotic variance no greater than that of the unadjusted estimator. The asymptotic variance can be estimated conservatively...
Loading next page...
 
/lp/taylor-francis/bootstrapping-inference-of-average-treatment-effect-in-completely-8AkL1ZDX1C
Publisher
Taylor & Francis
Copyright
© 2021 International Biometric Society – Chinese Region
ISSN
2470-9379
eISSN
2470-9360
DOI
10.1080/24709360.2021.1898269
Publisher site
See Article on Publisher Site

Abstract

Investigators often use regression adjustment methods to analyze the results of randomized experiments when baseline covariates are available. Their aim is to improve the estimation efficiency of treatment effects by adjusting for imbalance of covariates. Under mild conditions, the regression-adjusted average treatment effect estimator is asymptotically normal with asymptotic variance no greater than that of the unadjusted estimator. The asymptotic variance can be estimated conservatively based on residual sum of squares. This article studies alternative inference methods based on the bootstrap and investigates their asymptotic properties under the Neyman–Rubin causal model and randomization-based inference framework. We show that the weighted, residual and paired bootstrap methods provide asymptotically conservative variance estimators that perform at least as good as the estimator based on residual sum of squares. We further provide counterexamples, where the original estimator is asymptotically normal, but the bootstrap counterpart is inconsistent for estimating its limiting distribution. Simulation studies indicate that the paired bootstrap method is preferable, in terms of preserving type I errors, for a small sample size. Finally, our methods analyze HER2+ breast cancer data from the NeOAdjuvant Herceptin trial to examine the effectiveness of trastuzumab in combination with neoadjuvant chemotherapy.

Journal

Biostatistics & Epidemiology – Taylor & Francis

Published: Mar 17, 2021

Keywords: Causal inference; lasso; elastic net; bootstrap; penalized regression

References

  • On the application of probability theory to agricultural experiments. essay on principles (with discussion). section 9 (translated). reprinted ed
  • Estimating causal effects of treatments in randomized and nonrandomized studies
  • On regression adjustments to experimental data
  • On regression adjustments in experiments with several treatments
  • Agnostic notes on regression adjustments to experimental data: reexamining freedman's critique
  • Lasso adjustments of treatment effect estimates in randomized experiments
  • Causal inference from factorial designs by using potential outcomes
  • A bayesian perspective on the analysis of unreplicated factorial experiments using potential outcomes
  • On randomization-based and regression-based inferences for 2k factorial designs
  • Randomization-based causal inference from split-plot designs
  • Ridge regression: biased estimation for nonorthogonal problems
  • Regularization and variable selection via the elastic net
  • Adaptive lasso for sparse high-dimensional regression models
  • Large sample properties of matching estimators for average treatment effects
  • Bootstrap methods: another look at the jackknife
  • Bootstrap and wild bootstrap for high dimensional linear models
  • Some asymptotic theory for the bootstrap
  • Jackknife, bootstrap, and other resampling methods in regression analysis
  • A leisurely look at the bootstrap, the jackknife and cross-validation
  • Neoadjuvant chemotherapy with trastuzumab followed by adjuvant trastuzumab versus neoadjuvant chemotherapy alone, in patients with her2-positive locally advanced breast cancer (the noah trial): a randomised controlled superiority trial with a parallel her2-negative cohort
  • The randomization theory of experimental inference
  • General forms of finite population central limit theorems with applications to causal inference
  • Regression shrinkage and selection via the lasso
  • Confidence intervals for low dimensional parameters in high dimensional linear models
  • High-dimensional simultaneous inference with the bootstrap
  • Limiting distributions in simple random sampling from a finite population
  • Bootstrap inference of matching estimators for average treatment effects
  • The bayesian bootstrap
  • Regression adjustment for treatment effect with multicollinearity in high dimensions
  • Adjuvant docetaxel or vinorelbine with or without trastuzumab for breast cancer
  • Gene expression omnibus: ncbi gene expression and hybridization array data repository
  • Research-based PAM subtype predictor identifies higher responses and improved survival outcomes in HER -positive breast cancer in the NOAH study
  • Comparison of discrimination methods for the classification of tumors using gene expression data
  • Sure independence screening for ultrahigh dimensional feature space